ulrikpedersen
commented
1 year ago In the end I split the graylog configuration into a separate compose file. Then the logging services can be started independently of the simulation/control/daq services. The workflow to start up everything then becomes:
- First start logging services:
docker compose -f graylog-compose.yml up -d - Then start the controls environment:
docker compose up -d - Work on code, restart controls environment or individual services (
docker restart <container>) - and all service logs appears in graylog.
Adding Graylog services to docker compose and configuring the controls containers to redirect logs using the
GELFdriver.This approach just lumps the Graylog services (an extra mongo DB, elasticsearch and graylog containers) into the same docker compose file. If this becomes too unwieldy we could also consider splitting the graylog part into a separate compose file which can be kept running the in background as infrastructure... Thoughts, @8ryn ?
Closes #9